Bayes optimality in linear discriminant analysis

被引:106
作者
Hamsici, Onur C. [1 ]
Martinez, Aleix M. [1 ]
机构
[1] Ohio State Univ, Dept Elect & Comp Engn, Columbus, OH 43210 USA
关键词
linear discriminant analysis; feature extraction; Bayes optimal; convex optimization; pattern recognition; data mining; data visualization;
D O I
10.1109/TPAMI.2007.70717
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
We present an algorithm that provides the one-dimensional subspace, where the Bayes error is minimized for the C class problem with homoscedastic Gaussian distributions. Our main result shows that the set of possible one-dimensional spaces v, for which the order of the projected class means is identical, defines a convex region with associated convex Bayes error function g(v). This allows for the minimization of the error function using standard convex optimization algorithms. Our algorithm is then extended to the minimization of the Bayes error in the more general case of heteroscedastic distributions. This is done by means of an appropriate kernel mapping function. This result is further extended to obtain the d-dimensional solution for any given d by iteratively applying our algorithm to the null space of the (d - 1)-dimensional solution. We also show how this result can be used to improve upon the outcomes provided by existing algorithms and derive a low-computational cost, linear approximation. Extensive experimental validations are provided to demonstrate the use of these algorithms in classification, data analysis and visualization.
引用
收藏
页码:647 / 657
页数:11
相关论文
共 22 条
[1]  
[Anonymous], CLASSIFICATION CLUST
[2]  
[Anonymous], 1999, NEURAL NETWORKS SIGN
[3]  
[Anonymous], 1967, ANN MATH STAT
[4]  
[Anonymous], P IEEE C COMP VIS PA
[5]   Generalized discriminant analysis using a kernel approach [J].
Baudat, G ;
Anouar, FE .
NEURAL COMPUTATION, 2000, 12 (10) :2385-2404
[6]   A general model for finite-sample effects in training and testing of competing classifiers [J].
Beiden, SV ;
Maloof, MA ;
Wagner, RF .
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2003, 25 (12) :1561-1569
[7]   The statistical utilization of multiple measurements [J].
Fisher, RA .
ANNALS OF EUGENICS, 1938, 8 :376-386
[8]   NONPARAMETRIC DISCRIMINANT-ANALYSIS [J].
FUKUNAGA, K ;
MANTOCK, JM .
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 1983, 5 (06) :671-678
[9]  
Gill P.E, 1991, Numerical Linear Algebra and Optimization, V1
[10]  
LEIBE B, 2003, P IEEE C COMP VIS PA